| Abstract |
Smart city is a new urban governance model that combines technologies such as the Internet of Things, artificial intelligence, and 5G communication. It can process tourist traffic in real-time through a smart city system in the tourism field and provide technical support for monitoring regional abnormal behavior. In order to improve the safety detection effect of tourist attractions, this study proposes an intelligent abnormal behavior detection technology for scenic spots. The study first performs color adjustment and denoising preprocessing on scenic spot data. Secondly, the study proposes a detection model of hybrid multi-input feature clustering, which extracts the foreground through the target network and uses the interval frame sequence to represent the color difference, thereby realizing the extraction of multiple features, and uses the clustering algorithm to realize the classification of behavioral features. In the training loss analysis, the best model performance of the research model under the Avenue dataset is 0.254. In the training of the self-made dataset, the accuracy and detection time of the research model are the best, such as the accuracy rate is 0.998. It can be seen that research technology has excellent application effects and can provide technical support for the safety management of scenic spots. © COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only. |
